Factors Governing Oxygen Vacancy Formation in Oxide Perovskites

Abstract

The control of oxygen vacancy (V O ) formation is critical to advancing multiple metal-oxide-perovskite-based technologies. We report the construction of a compact linear model for the neutral V O formation energy in ABO 3 perovskites that reproduces, with reasonable fidelity, Hubbard-U-corrected density functional theory calculations based on the state-of-the-art, strongly constrained and appropriately normed exchange-correlation functional. We obtain a mean absolute error of 0.45 eV for perovskites stable at 298 K, an accuracy that holds across a large, electronically diverse set of ABO 3 perovskites. Our model considers perovskites containing alkaline-earth metals (Ca, Sr, and Ba) and lanthanides (La and Ce) on the A-site and 3d transition metals (Ti, V, Cr, Mn, Fe, Co, and Ni) on the B-site in six different crystal systems (cubic, tetragonal, orthorhombic, hexagonal, rhombohedral, and monoclinic) common to perovskites. Physically intuitive metrics easily extracted from existing experimental thermochemical data or via inexpensive quantum mechanical calculations, including crystal bond dissociation energies and (solid phase) reduction potentials, are key components of the model. Beyond validation of the model against known experimental trends in materials used in solid oxide fuel cells, the model yields new candidate perovskites not contained in our training data set, such as (Bi,Y)(Fe,Co)O 3 , which we predict may have favorable thermochemical water-splitting properties. The confluence of sufficient accuracy, efficiency, and interpretability afforded by our model not only facilitates high-throughput computational screening for any application that requires the precise control of V O concentrations but also provides a clear picture of the dominant physics governing V O formation in metaloxide perovskites.

Experimental protocols for the assessment of redox thermodynamics of nonstoichiometric oxides: A case study of YMnO3‐δ

“…Somewhat surprisingly, this value matches reasonably well atomistic level computational predictions of 335 kJ (mol‐O) −1 and 290 kJ (mol‐O) −1 made assuming a perovskite structure 24,25 . Ignoring the unusual behavior of the ferrielectric phase, the large enthalpy of reduction can be attributed to the large energetic penalty of changing the Mn oxidation state from 3+ to 2+ 26 . In the perovskite LaMnO 3 , for example, the enthalpy of reduction is 350 kJ (mol‐O) −1 , 27 similar to the value obtained here for YMnO 3 .…”

“…24,25 Ignoring the unusual behavior of the ferrielectric phase, the large enthalpy of reduction can be attributed to the large energetic penalty of changing the Mn oxidation state from 3+ to 2+. 26 In the perovskite LaMnO 3 , for example, the enthalpy of reduction is 350 kJ (mol-O) −1 , 27 similar to the value obtained here for YMnO 3 . In compounds such as CaMnO 3 , in which the reduction of Mn is from 4+ to a lower oxidation state, a lower enthalpy value has been measured, just 175 kJ (mol-O) −1 in the slightly oxygen deficient cubic phase.…”

Graph neural network modeling of vacancy formation enthalpy for materials discovery and its application in solar thermochemical water splitting

“…Due to the complexity of the associated vacancy defect calculations using DFT, approximately one year's work was required to build our training database (⇠ 200 host oxide compounds consisting of ⇠1500 unique defects). Our highly generalizable dGNN model then extends upon capabilities of previous work that requires carefully hand-engineered features in specific crystal systems (e.g., perovskites) 21,45 and obtains a similar expected mean absolute error in oxygen vacancy formation enthalpy (MAE < 450 meV) in an unseen compound (assuming its cations are represented in the training data). Depending on the model's defect predictions, oxide stability, and the stringency of these down-selection criteria, we then narrow down a small number of priority candidates for experimental efforts from 10,000s of possible MP oxides (among which are known STCH oxides we "re-discover" through our screening procedure).…”

“…40,63 For workable and economic thermodynamic conditions, these considerations lead to a desirable value for the oxygen vacancy defect formation energy in an interval of about [2.3, 4.0] eV. 21,41,58 Lower formation energies impair the ability to produce hydrogen in the oxidation step, while higher energies prevent significant changes of the O stoichiometry in the reduction step. To predict the defect formation energies for these "unseen" materials, we utilize the expectation across Kfold models, hD ĤY…”

“…Previous efforts to predict vacancy formation enthalpies span a Sandia National Laboratories, Livermore, California 94551, United States; E-mail: mwitman@sandia.gov, amcdani@sandia.gov various methods and material classes within which the models are applicable. Notable examples include modeling vacancy formation enthalpy using a simple hand-derived or machine learning (ML) model based on hypothesized important features [17][18][19][20][21] and descriptor derived properties to train ML regression models for defect property prediction in semiconductors 22,23 . For 2D materials consisting of TMDs, hexagonal boron nitride, and other selected wide band gap 2D materials, similar utilization of handengineered features and random forests predicted vacancy defect formation enthalpies.…”